Towards Standardization of Measuring the Usability of Parallel Languages

摘要

The efforts of the research community and the software industry to make the art of parallel programming easier continue. Measuring the usability of contemporary parallel programming languages and libraries by empirical studies is the key to understanding how programmers are thinking, designing, coding, and debugging parallel programs. In this paper we take apart into their component ingredients the empirical experiments done in the recent years. By analyzing each component separately we can better understand what is missing in these experiments and thereby improve the outcome of future studies. The result of this work is a set of recommendations that aims to make usability studies more convincing so that parallel language designers will take them seriously.